autoconf
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: AC_PROG_CC not working


From: Pavel Roskin
Subject: Re: AC_PROG_CC not working
Date: Tue, 10 Oct 2000 10:32:08 -0400 (EDT)

Hello, Akim and others!

> >> Anyway, I am still against the possibility to specify a list of
> >> compilers for AC_PROG_CC, CXX etc.  I think we will never stop
> >> having problem with this feature :(

Don't be pessimistic. We haven't tried it hard yet.

> Peter> This feature is extremely useful, because the default list is
> Peter> pretty useless for some platforms and packages.

Peter, it is extremely important that you give some real examples here.
Do you mean generic purpose free software or some very specific and
possibly closed-source projects?

I don't understand what "some platforms" you mean. Do those platform have
problems with most GNU software or with those specific packages. In the
former case, can you fix the problem using environment variable,
config.site and similar per-site tweaks?

> Then it means the list is wrong.  It's Autoconf which has to be
> fixed.  It's the tests that must be more accurate.

This statement is based on some assumptions that are probably correct but
let's see what Peter will answer.

> Very architecture depend issues, such as find *the* compiler, will
> probably never be properly handled by Autoconf.  It is the installer
> which has the best knowledge, and it is her that should pass to
> configure the right values.

Let's say it differently. Autoconf cannot guess your intentions or the
policy of your company, but it can (or or should be able to) be told about
them using site-specific and package-specific tweaks.

The question is whether some package-specific tweaks can be dropped in
favor of site-specific tweaks.

> Again, something which is extremely important with configures is that
> they must behave the same way.  Having different flavors of AC_PROG_CC
> seems wrong to me because it means the installer can not depend on
> some expected behavior.

That's a good point, as well as everything else below this.

But still I don't feel good about dropping an existing feature.

Regards,
Pavel Roskin




reply via email to

[Prev in Thread] Current Thread [Next in Thread]